Theory and empirical evidence suggest there may be significant heterogeneity in the developmental course of adolescent dating violence that could limit the effectiveness of one-size-fits-all prevention programs. For example, for some subgroups of adolescents, dating violence may be symptomatic of a systematic pattern of dysfunction that is life-course persistent, whereas for others involvement in dating violence may be a singular problem that desists during late adolescence. Intervention strategies may be more effective if they are tailored to meet the needs of these different subgroups and target those subgroups most at risk. Yet, despite a compelling theoretical and empirical rationale for examining heterogeneity in the development of teen dating violence, few studies have done so. In fact, to date, dating violence behavior has been investigated largely via cross-sectional studies with methods that may obscure meaningful heterogeneity in developmental processes. The present application addresses this gap in the literature by examining dating violence from a developmental science perspective, which views contextual and individual influences on behavior as being inextricably linked over time, using a set of influential methods that are well-suited for examining complex interactions and heterogeneity in developmental processes, namely, person-centered analytic approaches (e.g., latent class analysis, growth mixture modeling). Person-centered methods may be used to identify distinct subgroups of individuals who are similar to each other within a group, but differ from members of other groups, in terms of their pattern of values across an entire set of variables. These patterns can indicate which types of adolescents are at greatest risk for dating violence or are in need a particular intervention. To this end, the specific aims of the proposed study are to use person-centered approaches to: (1) identify subgroups of middle school adolescents with distinct risk/protective profiles and examine the relation of these profiles to future dating violence; (2) identify subgroups of adolescents based on their patterns of involvement in different types (peer, dating) and forms (physical, psychological, sexual) of victimization and perpetration behavior and examine transitions between groups over time; and (3) identify and characterize qualitatively distinct groups of developmental trajectories of physical, psychological, and sexual dating violence perpetration across grades 8 through 12. Across all aims sex differences in subgroup structure and membership and in relations between subgroup membership and covariates will be examined. Study aims will be accomplished by applying a variety of different person-centered methods (e.g., latent class analysis, conjoint growth mixture modeling) to extant data from two multi-wave studies, enabling cross-validation of findings across data sources. Work under the proposed aims is expected to illuminate heterogeneous patterns of risk, co-morbidity and change in adolescent dating violence. Findings will inform our theoretical understanding of dating violence and contribute to identify subgroups of youth that may benefit from different kinds, doses, or levels of intervention.